- FanGraphs Baseball - http://www.fangraphs.com/blogs -

A Look at Workhorse Pitchers

Earlier this week, in the aftermath of the big Royals-Rays trade centered around Wil Myers and James Shields, Dave Cameron wrote about the perception of relative risk with respect to veterans and top prospects. One issue in this case is the supposed reliability of a “workhorse” pitcher like James Shields. Dave rightly pointed out that although prospects are risky, there is risk associated with any player, even a supposedly reliable pitcher. I wanted to look a bit more closely at these workhorse pitchers. After a number of seasons of 200 or more innings, how many innings do such pitchers put up in the subsequent seasons?

I should note at the outset that this is not an overly formal study. I think that the information presented here has value, but simply do not want to make exaggerated claims for its rigor or predictive value. For example, I made number of assumptions that are somewhat arbitrary and could probably be improved upon, although I did have my reasons for making them.

I make no claims for this study being revolutionary, nor do I expect any surprising results. It is a pretty safe assumption that if a pitcher has pitched a lot of innings for a few years, he will probably do so in the following year. However, it is often worth checking things out to see if those assumptions are as safe as we think. With that out of the way, let’s see what I found.

The first thing I wanted to do was get a rough definition of “workhorse,” or a pitcher who has pitched a consistently large number of innings for a few seasons in a row. This presents a number of issues. First of all, innings pitched is an imperfect measure. A pitcher might get a higher or lower number of innings from season to season based on a number of factors. It could be due to injury, of course. But the pitcher could also have the same raw ability to throw the same number of pitches, but is less efficient in getting outs per pitch. The pitcher’s ability (perceived or real) to prevent runs may be better or worse, and thus the pitcher gets moved up or down in the rotation, or is more or less likely to have his turn skipped. Other factors could be mentioned, but this gets the point across.

Despite those issues, innings pitched is commonly used, does not seem to be worse than other candidates, and its admittedly composite nature makes it helpful for a study like this, which is intended to be a generic one that does note try to isolate factors such as performance and injury.

Finding a definition of “workhorse pitcher” is another issue. My idea to start with pitchers who had pitched 200 or more innings three seasons in a row, for example, ran up against the obvious problem that pitchers used to pitch far more innings than they do in contemporary baseball. However, the bigger the sample, the more informative the study might be. Since 1955 (the conventional beginning of the Retrosheet Era), there have been 807 instances of three 200-plus innings seasons in a row. However, back in the day, this was not a huge deal. Wilbur Wood threw more than 300 innings for four seasons in a row from 1971 to 1974, and in 1975 he threw 291. That is probably worth a post in itself, but while the earlier eras can be helpful, for a simple study like this one that is concerned with contemporary starting pitchers, not so much.

As a somewhat arbitrary dividing line, I decided to use 1980 — the last year in which a pitcher (Steve Carlton) threw 300 or more innings. It is true that the top starting pitchers even in the 1980s threw more innings than the top starters today. As we will see, though, it does not make a huge difference for my purposes.

Let’s get to some actual data. I took all instances of pitchers pitching 200 or more innings three years in a row since 1981 and looked at what they did in the fourth season. Some seasons and pitchers can be used more than once, e.g., James Shields 2007-2009 and 2008-2010 both count as separate series of three seasons. I also excluded significantly strike-shortened seasons as possibilities for the fourth year (it did make a difference). I found 351 instances of this starting in 1981.

Since 1981 there have been 351 cases of three straight 200+ inning seasons. In the fourth year, those pitchers pitched an average of 86.1 percent their average number of innings for the previous three years. In other words, if a pitcher pitched 600 innings over three seasons, in fourth season, if he would have been average in this respect, he would have pitched about 172 innings.

Does it make a big difference if restrict our sample to a more recent time period? This is worth checking out because top starting pitchers are pitching fewer innings than they were 20 or 30 years ago. Since 1991, there have been 182 cases of 200 or more innings pitched over a season. In the fourth year, those pitchers averaged 87.1 percent of the average of their three previous seasons. Since 2001, there have been 106 cases, and the fourth-year average has been about 88.7 percent of the average innings pitched from the three prior years. There has been some change, but it does not seem to be terribly dramatic, so we can use the larger sample without too much concern.

Of those 351 cases, in 93 (about 26 percent) of them the pitchers increased their fourth year innings over their previous three-year average, although that is not surprising given what the average turned out to be. The majority, obviously, decreased, but how risky is that group?

Of the 258 who decreased in year four, 167 (about 65 percent) pitched at least 80 percent of their prior-three-year average, 63 (about 24 percent) pitched between 50 and eighty percent of that average. Of the total group of 258, only 28 (about 11 percent) pitched less than half of their three-year average innings in the fourth year. One could break this down further, but the story seems pretty clear: even of the decliners, more than half (percent) retain at least 80 percent of their previous level of endurance, and only about one in ten see a dramatic collapse to less than half of their previous average.

Another interesting way of looking at this would be to see if there is much of a difference by age. Here is that average fourth year average percent change from the previous three-year average grouped by age, where n is the number of instances:

4th age n 4th
22 1 0.7224
24 1 1.0166
25 8 1.0391
26 21 0.8675
27 32 0.9017
28 31 0.9307
29 44 0.889
30 34 0.8663
31 31 0.8309
32 30 0.8885
33 28 0.8173
34 15 0.8728
35 15 0.7993
36 15 0.7615
37 9 0.8952
38 8 0.8095
39 3 0.72
40 6 0.8431
41 8 0.6585
42 2 0.908
43 4 0.6806
44 2 0.8785
46 1 1.0146
47 1 0.9904
48 1 0.6425

Obviously, at the extreme ends the sample is so small as to not be useful: on one end we have Doc Gooden as the whole sample, on the other (late thirties and forties) we have Phil Nierko. The sample is not exactly giant in the age-ranges from the late twenties to the early thirties, but at least there is more there. We basically see the sorts of things we would expect: players with their fourth year coming prior to 30 see a smaller drop-off in innings than those seeing their fourth year after 30. Starting at about 29, the drop-off from the three-year average is usually between 10 and 15 percent, and gets closer to twenty percent as the thirties progress. Be careful not to taken in by the shrinking sample sizes and selection bias in the late thirties data!

One thing to remember in this admittedly limited study is that it should not be used to say anything about pitcher aging or endurance in general, as it deliberately selected pitchers who have pitched large numbers of innings, and were allowed to do so for a reason. However, it might tell us something about what we can expect from “workhorse” pitchers and how many innings they can be expected to pitch.

Even pitchers who have consistently put up 200 or more innings for three seasons in a have pitched 10 to 15 percent fewer than the average of those three years in the fourth year. Despite that drop-off, we have also found it to be rare that they have a collapse — whether due to health or ineffectiveness — that causes them to pitch far less than they had in the previous three years. About two-thirds of these pitchers pitch at last eighty percent of average innings from the previous years, and only about one-tenth pitched less than half of the innings. In other words, counting on that 200-plus inning pitcher to do so year after year is overly optimistic. However, while there is a fair amount of risk, chances are that the workhorse is not going to fall apart, either. And, unsurprisingly, the younger the pitcher is, the better.

As a final note, I want to try and narrow down the population a bit more to see if we can find something more specific to James Shields, a big part of the original inspiration for this work. After all, Shields has put in six seasons in a row of 200-plus innings. Since 1981, there have been 64 instances where a pitcher pitched at least 200 innings six seasons in a row and come back for a seventh. In that seventh seasons, they have pitched an an average of 88 percent of their average seasonal innings from the previous six years. That is a bit better, but not all that different from the three-year instances.

Naturally, this is just a general study. For predicting the durability and effectiveness of any particular pitcher, one would want to look at specifics such as velocity, repertoire, and so on. But just looking at the overall population of consistent workhorses from the past 30 years, while Shields and his contemporary brethren should be expected to pitch fewer innings as a group, they are unlikely to collapse en masse, either.